Can Magnetic Resonance Imaging-Derived Bone Models Be Used for Accurate Motion Measurement with Single-Plane Three-Dimensional Shape Registration?

The purpose of this study was to compare three-dimensional (3D) kinematic measurementsfromsingle-planeradiographicprojectionsusingbonemodelscreatedfrommagnetic resonance imaging (MRI) and computed tomography (CT). MRI is attractive because there is no ionizing radiation, but geometric field distortion and poor bone contrast degrade model fidelity comparedtoCT.WecreatedkneebonemodelsofthreehealthyvolunteersfrombothMRIandCTand performed three quantitative comparisons. First, differences between MRI- and CT-derived bone modelsurfacesweremeasured.Second,shapematchingmotionmeasurementsweredonewithbone models for X-ray image sequences of a squat activity. Third, synthetic X-ray images in known poses were created and shape matching was again performed. Differences in kinematic results were quantified in terms of root mean square (RMS) error. Mean differences between CT and MRI model surfaces for the femur and tibia were � 0.08 mm and � 0.14 mm, respectively. There were significant differences in three of six kinematic parameters comparing matching results from MRI-derived bone models and CT-derived bone models. RMS errors for tibiofemoral poses averaged 0.74 mm for